Page adaptive construction method and device based on cognitive state transition and storage medium

Abstract

The application discloses a page adaptive construction method and device based on cognitive state transition and a storage medium, and the method comprises the following steps: modeling the generation process of an interactive teaching page as a cognitive state transition network; obtaining a cognitive transition cost by quantifying the cognitive jump difficulty between adjacent cognitive state nodes, determining an optimal teaching path from an initial cognitive state to a target cognitive state based on the minimum principle of the cognitive transition cost; determining a teaching section sequence according to the optimal teaching path, each teaching section corresponding to a cognitive state node, mapping each teaching section to a page container based on a mapping system from an educational cognitive theory to a page structure, generating response content for different page containers, establishing a section linkage relationship, and outputting a teaching page. The application effectively avoids understanding interruption caused by cognitive cliff, and significantly improves the cognitive smoothness and learning effectiveness of the teaching page.

Description

Technical Field

[0001] This invention belongs to the technical fields of educational technology, intelligent teaching content generation, and cognitive computing, and specifically relates to a page adaptive construction method, device, and storage medium based on cognitive state transition. Background Technology

[0002] With the development of large language models, multimodal content generation technology, and front-end interaction technology, the automatic generation of teaching content using intelligent systems has become an important development direction in the field of educational technology. Existing technologies can usually generate explanatory texts, exercise content, courseware fragments, or web pages based on input topics, but most of them treat teaching content generation as a general text generation task or a general page generation task, focusing more on the completeness of the generated content or the displayability of the page itself, while lacking a structured organization of the teaching process.

[0003] Existing adaptive learning systems mainly focus on question recommendation and difficulty matching. For example, US patent US10943497B2 proposes to use a deep knowledge tracing (DKT) model to determine the next question and predict the learner's future performance. US patent US11631338B2 further uses the transformer framework for knowledge tracing. Chinese patent CN114201684A combines a cognitive diagnostic model, a test score prediction model, and a knowledge graph to recommend learning resources. Chinese patent CN118364904A directly predicts the student's performance on the next question for the knowledge tracing task. However, these systems have the following limitations: First, they focus on "what to choose in the next question" rather than "how to design the next page," simplifying teaching into discrete knowledge point jumps and ignoring the continuity of the teaching process; second, they lack explicit modeling of cognitive load fluctuations, which can easily cause learners to experience cognitive overload or under-cognition when the difficulty of the questions recommended by the system jumps too much; third, they do not incorporate the leaps in human cognitive levels into the optimization objectives, which may result in cognitive cliffs in the recommended exercise sequences (such as jumping directly from the "understanding" level to the "creation" level).

[0004] In the field of intelligent courseware generation, existing solutions are capable of generating courseware pages or instructional designs. For example, WO2020211397A1 involves the display of courseware pages and the construction of page sets, CN117994101B involves instructional design generation based on large language models, and CN113468288A involves text courseware content extraction and semantic structure processing based on artificial intelligence. Existing solutions can generate courseware pages containing text, illustrations, and animations based on teaching themes, but they typically use static template filling, resulting in a fixed page structure. These solutions fail to establish a dynamic mapping relationship between educational cognitive theory and page structure, and cannot adjust the page segment sequence in real time according to the learner's cognitive state. More importantly, existing solutions treat the page structure as a pre-set fixed set, rather than a dynamic sequence evolved from cognitive states, making it difficult for the page to adapt to the cognitive trajectories of different learners.

[0005] In the field of generative UI technology, solutions that dynamically combine large language models with front-end components have recently emerged. For example, Stanford University's GenUI framework (Generative Interfaces for Language Models) can dynamically predict and render appropriate UI components (such as forms and charts) based on the dialogue context to replace plain text; another example is Google's A2UI (Agent-to-UI) protocol, which supports agents streaming interactive UI components to the front end and can be applied to personalized learning scenarios. However, a key flaw in existing technologies is that these cutting-edge generative UI solutions (such as GenUI and A2UI) are essentially still "reactive mapping"—they merely predict and render the "next most suitable UI component" in real time based on the user's current single-step input or prompt. In educational scenarios, this "step-by-step" generation, lacking global cognitive planning, has a fatal weakness:

[0006] First, they lack explicit and continuous modeling of learners’ deep cognitive states (such as Bloom’s hierarchy of cognition and cognitive load). The generation of UI is limited to the local dialogue context, rather than the cognitive evolution trajectory in the educational sense.

[0007] Second, they failed to model the generation of page component sequences as a globally optimal path problem in the cognitive state space. For example, although A2UI's personalized learning application can push practice question components, it cannot calculate the cognitive transition cost between concept explanations and practice questions, which can easily lead to cognitive cliffs between consecutively delivered components.

[0008] Third, they view the teaching process as a linear, fluid stacking of UI components, rather than an ordered evolution of cognitive states. Teaching pages should not merely be a combination of controls rendered according to intent, but rather a sequence of teaching containers evolving from a path that minimizes cognitive cost.

[0009] This deficiency means that while existing generative UI technologies achieve dynamic and interactive interfaces in form, they fall short in terms of cognitive smoothness, instructional coherence, and overall learning effectiveness. This invention aims to overcome these cognitive blind spots in the application of generative UIs in education.

[0010] Furthermore, the structure of instructional pages should vary depending on different learning objectives and learner conditions. For example, for learners with weaker foundations, pages typically require more examples, stronger guidance, and lower interaction complexity; while for learners with some foundation, pages usually require richer variable operations, tighter comparison structures, and a greater degree of independent exploration. Current technology lacks a solution for organizing and adjusting instructional page objects, making it difficult to effectively translate learner characteristics into differences in page structure, section order, content density, and control complexity. It also makes it difficult to reasonably control page information density and interaction complexity based on learners' working memory capacity limitations, easily leading to information overload or insufficient cognitive load.

[0011] Therefore, the inventors propose an interactive teaching page construction method based on a teaching plan, so that the automatic construction result no longer stops at the output of general content or general web pages, but forms an executable teaching page composed of multiple page segments with teaching functions. Through the segment content writing and interactive linkage mechanism, the page can carry a continuous teaching process, thereby improving the classroom usability, learning adaptability and interactive effectiveness of automatically generated teaching content. Summary of the Invention

[0012] The main objective of this invention is to overcome the shortcomings and deficiencies of the prior art and provide a page adaptive construction method, device and storage medium based on cognitive state transition. By abstracting the teaching process into a cognitive state sequence and transforming page construction into an optimal path solving problem, this invention realizes a paradigm shift in teaching pages from "design-driven" to "cognitive evolution-driven". This enables the page structure to be automatically optimized according to the cognitive transition cost, effectively avoiding comprehension interruptions caused by cognitive cliffs, and significantly improving the cognitive smoothness and learning effectiveness of teaching pages.

[0013] To achieve the above objectives, the present invention adopts the following technical solution:

[0014] In a first aspect, the present invention provides a page adaptive construction method based on cognitive state transition, comprising the following steps:

[0015] The generation process of interactive teaching pages is modeled as a cognitive state transition network, where each cognitive state node represents a cognitive state at a certain moment in the teaching process.

[0016] The cognitive transition cost is obtained by quantifying the difficulty of cognitive jumps between adjacent cognitive state nodes. Based on the principle of minimizing cognitive transition cost, the optimal teaching path from the initial cognitive state to the target cognitive state is determined.

[0017] The teaching segment sequence is determined based on the optimal teaching path. Each teaching segment corresponds to a cognitive state node. Based on the mapping system from educational cognition theory to page structure, each teaching segment is mapped to a page container. Response content is generated for different page containers, and the segment linkage relationship is established to output the teaching page.

[0018] As a preferred technical solution, the process of generating interactive teaching pages is modeled as a cognitive state transition network, specifically as follows:

[0019] It receives teaching topics, teaching materials, and learner parameters, and forms an input data package;

[0020] Perform missing parameter completion and standardization on the learner parameters to obtain a computable set of learner parameters. The learner parameter set is represented as follows:

[0021] ,

[0022] in, Indicates the learner's prior level. Indicates the learner's level of comprehension. Indicates the degree of interaction required. Indicates the required intensity of guidance;

[0023] The large language model is invoked to perform semantic parsing and structured extraction on the teaching topics and teaching materials. Based on the semantic parsing results of the teaching topics, the knowledge scope covered by the interactive teaching page, the knowledge units to be generated, and the target cognitive level corresponding to each knowledge unit are determined. Based on the semantic parsing results of the teaching materials, teaching content resources associated with each knowledge unit are generated, and the prerequisite relationships between knowledge units and the executable teaching actions corresponding to each knowledge unit are extracted. Cognitive state nodes are defined by combining each knowledge unit, the corresponding cognitive level, the target load of the state layer, and the learner's mastery level.

[0024] Based on at least one of the learner's historical responses, historical interaction behaviors, dwell time, number of repeated operations, and feedback results for each knowledge unit, the learner's mastery level of each knowledge unit is determined; when the historical learning data is missing, the learner's prior level in the learner parameters pre-input by the learner is obtained, the learner's prior level is processed for missing information completion and standardization, and the learner's initial mastery level of each knowledge unit is determined according to the preset mastery level mapping rules;

[0025] Cognitive state nodes are defined by combining knowledge units, cognitive levels, state-level target load, and mastery levels. As shown in the following formula:

[0026] ,

[0027] in, Represents a knowledge unit. Indicates cognitive level, Represents cognitive state nodes The corresponding state layer target load, This indicates the learner's level of mastery of the corresponding knowledge unit;

[0028] The executable teaching actions corresponding to the transition between cognitive state nodes are used as state transition edges. The range of state transition edges that can be established is limited according to the prior relationships between the knowledge units, and a set of state transition edges is obtained.

[0029] Based on the learner's prior knowledge level, the learner's acceptance level, and the learner's mastery of the corresponding knowledge unit, the initial cognitive state node is determined from the cognitive state nodes; based on the target knowledge unit, target cognitive level, and preset target mastery level corresponding to the teaching topic, the target cognitive state node is determined.

[0030] Based on the set of cognitive state nodes and the set of state transition edges, a cognitive state transition network is constructed.

[0031] As a preferred technical solution, the cognitive transition cost is obtained by quantifying the difficulty of cognitive jumps between adjacent cognitive state nodes. Specifically, the cognitive transition cost function is defined as follows:

[0032] ,

[0033] in, Indicates the magnitude of leaps in cognitive levels. This represents the change in target load at the state level. This represents an indicative function, indicating when a cognitive level jump exceeds a threshold. Punishment will be imposed at that time.

[0034] As a preferred technical solution, the optimization objectives of the teaching path include minimizing the total transition cost, minimizing cognitive load fluctuations, and minimizing the number of steps in the path.

[0035] The total transition cost, cognitive load fluctuation, and minimum path steps are weighted and summed according to preset weights to obtain the comprehensive path cost corresponding to the candidate teaching path.

[0036] The minimized total transition cost is used to reduce the total number of cognitive jumps between adjacent cognitive state nodes in the entire teaching path; the minimized cognitive load fluctuation is used to reduce the fluctuation of the state layer target load of each cognitive state node in the teaching path; and the minimized path steps are used to control the length of the main path of the page.

[0037] For candidate teaching paths:

[0038] ,

[0039] Its integrated path cost function is:

[0040] ,

[0041] The optimal teaching path is determined according to the following formula:

[0042] ,

[0043] in, This represents the i-th cognitive state node. Represents cognitive state nodes The corresponding state layer target load, where T represents the number of path steps in the candidate teaching path. These are used to adjust the influence of total transition cost, cognitive load fluctuation, and path steps on path selection, respectively. This represents the set of candidate teaching paths that satisfy the teaching constraints. CTC (·) represents the cognitive transition cost function. Var (·) represents the cognitive load fluctuation function;

[0044] The teaching constraints include at least the prerequisite dependency constraints between knowledge units, the cognitive level transfer constraints, the state layer target load threshold constraints, the accessibility constraints, and the maximum number of steps on the main page path constraints.

[0045] The prerequisite dependency constraint is used to limit the cognitive state node corresponding to the prerequisite knowledge unit to be located before the cognitive state node corresponding to the target knowledge unit when the learner's mastery of the prerequisite knowledge unit is lower than a preset mastery threshold; when the learner's mastery of the prerequisite knowledge unit reaches the preset mastery threshold, the candidate teaching path is allowed not to include the cognitive state node corresponding to the prerequisite knowledge unit.

[0046] The cognitive level transfer constraint is used to determine, based on a preset cognitive level transfer matrix, whether a transfer between cognitive levels corresponding to adjacent cognitive state nodes is allowed, and satisfies the following:

[0047] ,

[0048] in, This represents the preset cognitive level transition matrix. and Representing adjacent cognitive state nodes respectively and The corresponding cognitive level, with a matrix element value of 1 indicating that transitions between the corresponding cognitive levels are allowed;

[0049] The state-level target load threshold constraint is used to limit the state-level target load corresponding to each cognitive state node in the candidate teaching path from not exceeding the state-level target load threshold determined according to the learner parameters, and satisfies:

[0050] ,

[0051] in, Represents cognitive state nodes The corresponding state layer target load, Indicates based on learner parameters The maximum target load at the determined state level;

[0052] The reachability constraint is used to limit the candidate teaching path to start from the initial cognitive state node and end at the target cognitive state node, and to ensure that there is a corresponding state transition edge and an executable teaching action between any two adjacent cognitive state nodes; the maximum number of steps in the main page path constraint is used to limit the number of steps in the candidate teaching path to not exceed a preset maximum number of steps, and to satisfy:

[0053] ,

[0054] ,

[0055] in, This represents the initial cognitive state node. Represents the target cognitive state node. This represents the set of state transition edges in a cognitive state transition network. Represents cognitive state nodes Transfer to cognitive state node The corresponding executable teaching actions, This represents the number of steps in a candidate teaching path. This indicates the maximum number of steps in the preset main path of the page.

[0056] As a preferred technical solution, the step of determining the sequence of teaching segments based on the optimal teaching path, where each teaching segment corresponds to a cognitive state node, is based on a mapping system from educational cognition theory to page structure, specifically as follows:

[0057] For each cognitive state node in the optimal teaching path Based on the cognitive level corresponding to the cognitive state node, the corresponding basic segment type is determined through a preset segment type mapping relationship. Based on the arrangement order of the cognitive state nodes in the optimal teaching path, a basic segment type sequence is formed:

[0058] ,

[0059] According to the aforementioned basic segment type sequence, each basic segment type and their corresponding cognitive state nodes As input, generate basic page segment skeletons that correspond one-to-one with each cognitive state node. The basic page segment skeleton is a graphical user interface segment object to be instantiated, including a segment type identifier, a binding relationship with cognitive state nodes, and content slots, interaction slots, and feedback slots for carrying teaching content, interactive controls, and feedback information; wherein, the basic segment type is used to determine the teaching function of the basic page segment skeleton, and the cognitive state nodes are used to determine the knowledge unit, cognitive level, state layer target load, and mastery level constraints corresponding to the basic page segment skeleton;

[0060] Based on the knowledge type corresponding to each knowledge unit, the corresponding basic page section skeleton is... Enhance at least one of the internal content structure, interaction structure, and feedback structure to obtain an enhanced page segment skeleton. The knowledge types are identified by the large language model based on the teaching theme and teaching materials, including declarative knowledge, procedural knowledge, and conditional knowledge; the enhancement process maintains the order and number of cognitive state nodes in the optimal teaching path, and the enhanced page segment skeleton... The number corresponds to the number of cognitive state nodes in the optimal teaching path;

[0061] Enhance the page section skeleton Assign structure parameters, define the first The segment parameter vector corresponding to the skeleton of each enhanced page segment:

[0062] ,

[0063] in, Indicates the number of information elements. Indicates content density, Indicates the interaction complexity. Indicates feedback density;

[0064] The segment parameter vector is converted into a computable cognitive load value, the first... The actual cognitive load of an enhanced page section skeleton is:

[0065] ,

[0066] in, , and The weighting coefficients represent the degree of influence of the number of information elements, interaction complexity, content density, and feedback density on the cognitive load of the segment, respectively; learner parameters are defined. As shown in the following formula:

[0067] ,

[0068] in, Indicates the learner's prior level. Indicates the learner's level of comprehension. Indicates the degree of interaction required. Indicates the required intensity of guidance;

[0069] The segment structure parameters are adjusted based on the learner parameters, as shown in the following formula:

[0070] ,

[0071] ,

[0072] ,

[0073] ,

[0074] in, This represents the basic content density of the i-th enhanced page segment skeleton. Indicates the complexity of basic interactions. Indicates the basic feedback density. This indicates the number of basic information elements in the skeleton of the i-th enhanced page segment. and These represent the scaling adjustment functions for the corresponding learner parameters, and their range or output results are determined by system preset rules, teacher configuration, or historical teaching data.

[0075] Based on the enhanced page segment skeleton Determine the segment parameter vector As shown in the following formula:

[0076] ,

[0077] in, Represents the function for solving segment parameters. Represents cognitive state nodes The corresponding state layer target load and learner parameters The segment parameter vector is determined by the segment parameter solution function. ;

[0078] The determined segment parameter vector Write the corresponding enhanced page section skeleton Based on the teaching content, content structure, interaction structure, feedback structure, and layout rules in the enhanced page segment skeleton, as well as the number of information elements, content density, interaction complexity, and feedback density limited by the segment parameter vector, the enhanced page segment skeleton is instantiated to obtain renderable and interactive page segments. As shown in the following formula:

[0079] ,

[0080] in, This function instantiates a page segment, which is used to write teaching content into the enhanced page segment skeleton, generate corresponding interactive controls and feedback components, and determine the page layout based on the objects inside the segment and their relationships.

[0081] As a preferred technical solution, mapping each teaching segment to a page container specifically involves:

[0082] Based on the enhanced page segment skeleton Determine the teaching function of each section and generate the corresponding internal objects;

[0083] The internal object structure corresponding to the segment is determined based on the cognitive state node, including cognitive level, target load and mastery level constraints, and the problem situation or task background of the segment is adjusted using the mastery level constraints.

[0084] Based on the segment parameter vector Determine the page variables within the section, including the number of information elements, content density, interaction complexity, and feedback density;

[0085] Automatically select interactive controls based on the characteristics of page variables, including: slider controls for continuous variables, selection controls for discrete variables, input box controls for text input, and question and submission components for true / false or multiple-choice questions.

[0086] The page layout is determined based on the number, type, and relationships of internal objects, and the teaching content, interactive controls, and feedback components are written into the corresponding content slots, interaction slots, and feedback slots to generate the renderable and interactive page segments. .

[0087] As a preferred technical solution, the establishment of segment linkage includes:

[0088] Establish a linkage between the interactive triggering section and the update target section, so that learners can trigger content updates across sections when they operate in a single section;

[0089] When the same operation triggers synchronous responses in multiple segments, updates are performed according to preset linkage priorities, which are observation, comparison, summarization, and verification from high to low.

[0090] Secondly, the present invention also provides a page adaptive construction system based on cognitive state transition, applied to the aforementioned page adaptive construction method based on cognitive state transition, including an input module, a state network construction module, a path optimization module, a segment mapping module, a parameter adjustment module, a content generation and writing module, a linkage establishment module, and an output module;

[0091] The input module is used to receive teaching topics, teaching materials, and learner parameters, and form an input data package;

[0092] The network construction module is used to model the generation process of interactive teaching pages as a cognitive state transition network, wherein each cognitive state node in the cognitive state transition network represents the cognitive state at a certain moment in the teaching process.

[0093] The path optimization module is used to obtain the cognitive transition cost by quantifying the cognitive jump difficulty between adjacent cognitive state nodes, and to determine the optimal teaching path from the initial cognitive state to the target cognitive state based on the principle of minimizing cognitive transition cost.

[0094] The segment mapping module is used to determine the sequence of teaching segments based on the optimal teaching path. Each teaching segment corresponds to a cognitive state node, which is a mapping system based on educational cognition theory to page structure.

[0095] The parameter adjustment module is used by the system to calculate the segment parameter vector for each teaching segment after determining the corrected segment type, adjust the segment parameters, and determine the segment parameter vector.

[0096] The content generation and writing module is used to map each teaching section into a page container;

[0097] The linkage establishment module is used to establish the linkage relationship between the interactive triggering section and the update target section, so that the learner's operation in a single section can trigger content updates across sections.

[0098] The output module is used to output the teaching pages.

[0099] Thirdly, the present invention provides an electronic device, the electronic device comprising:

[0100] At least one processor; and,

[0101] A memory communicatively connected to the at least one processor; wherein the memory stores computer program instructions executable by the at least one processor, the computer program instructions being executed by the at least one processor to enable the at least one processor to execute the page adaptive construction method based on cognitive state transition.

[0102] Fourthly, the present invention provides a computer-readable storage medium storing a program, which, when executed by a processor, implements the page adaptive construction method based on cognitive state transition.

[0103] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0104] (1) This invention realizes the transformation of generative UI to global, effectively eliminating the "cognitive cliff" phenomenon in the dynamic component distribution. Breaking through the limitation of existing cutting-edge solutions that can only passively respond to local contexts, this invention gives the page generation process a global evolutionary perspective and rigorous educational theory support. The system can adaptively find the best teaching balance between "smooth but lengthy" and "compact but jumpy". Compared with the traditional generation sequence that does not adopt the cognitive state transfer method, this invention can effectively reduce the fluctuation of learners' cognitive load, reduce the occurrence of large-span cognitive jumps, improve the continuity of interaction, and reduce the abandonment rate of the page in the middle.

[0105] (2) This invention breaks through the shallow content replacement of traditional adaptive learning and realizes a deep personalized reorganization of page structure and information density. This invention transforms the working memory capacity limit into a computable machine constraint, so that the same teaching topic can automatically reorganize and generate completely different page container sequences, interaction complexity and feedback mechanisms when facing different learner characteristics. This not only achieves precise error prevention against information overload, but also makes the generation of the interface and the control of difficulty have a strict quantitative cognitive science basis, maximizing the inclusiveness of teaching and the adaptability of learners.

[0106] (3) This invention constructs a highly fault-tolerant organic interactive feedback chain, enabling the generated interactive teaching page to have a complete structure that can be directly executed. This invention effectively reduces the probability of problems such as mismatched control types and incompatible layouts in the automatically generated interface. At the same time, the interactive linkage mechanism between sections enables learners' operations to generate collaborative feedback between multiple sections, reducing the isolated interaction problem caused by the stacking of static components. Finally, it outputs a complete and fully functional executable page, reducing the cost of manual integration in the later stage.

[0107] (4) This invention realizes a paradigm shift in the construction of teaching pages: from "preset template filling" to "cognitive state evolution driven", from "fixed segment set" to "dynamic generation of optimal path", and from "qualitative experience rules" to "quantitative leap cost optimization". By abstracting the teaching process into a cognitive state transition network, this invention fundamentally solves the problems of the disconnect between page structure and cognitive evolution, lack of theoretical constraints on teaching sequence, and insufficient learner adaptability in the prior art, and significantly improves the cognitive smoothness, teaching effectiveness and learning adaptability of automatically generated teaching pages. Attached Figure Description

[0108] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0109] Figure 1 This is a flowchart of the page adaptive construction method based on cognitive state transition according to an embodiment of the present invention;

[0110] Figure 2 This is a schematic diagram of the cognitive state transition network and optimal teaching path according to an embodiment of the present invention;

[0111] Figure 3 This is a schematic diagram illustrating the generation and instantiation of the cognitive state node to the teaching page segment in an embodiment of the present invention;

[0112] Figure 4 This is a schematic diagram of the teaching page skeleton and segment container organization according to an embodiment of the present invention;

[0113] Figure 5 This is a schematic diagram of the inter-segment linkage relationship in an embodiment of the present invention;

[0114] Figure 6 This is a schematic diagram of the system structure of the page adaptive construction system based on cognitive state transition according to an embodiment of the present invention;

[0115] Figure 7 This is a structural diagram of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0116] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of the present application, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present application without creative effort are within the scope of protection of the present application.

[0117] In this application, the reference to "embodiment" means that a specific feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a mutually exclusive, independent, or alternative embodiment. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described in this application can be combined with other embodiments.

[0118] The terminology explained in this embodiment is as follows:

[0119] Dijkstra's algorithm, also known as the "Dijkstra's Shortest Path Algorithm," uses cognitive states as nodes in a graph, transition relationships between cognitive states as edges, and the cost of transitions between cognitive states as edge weights. Dijkstra's Shortest Path Algorithm calculates the learning path with the minimum cumulative transition cost to reach the target cognitive state, starting from the learner's current cognitive state.

[0120] Please see Figure 1 This embodiment provides a method for adaptively constructing interactive teaching pages based on cognitive state transitions. Instead of directly generating static pages based on the teaching topic, this method first constructs a cognitive state transition network based on learner parameters and teaching objectives. Then, it solves for the optimal teaching path in the cognitive state space. Subsequently, the optimal teaching path is converted into a sequence of segment containers on the teaching page, and further, segment content, interactive controls, layout structure, and inter-segment linkage relationships are generated. The method includes the following steps.

[0121] S1. The generation process of the interactive teaching page is modeled as a cognitive state transition network, where each cognitive state node in the cognitive state transition network represents the cognitive state of the corresponding knowledge unit at a certain stage in the teaching process.

[0122] Receive teaching topics, teaching materials, and learner parameters to form an input data packet. .

[0123] in, This indicates the teaching topic and is used to determine the scope of knowledge covered in this interactive teaching page; Teaching materials can include text materials, images, examples, questions, experimental data, animation clips, or other resources that can be used to generate teaching content; This indicates learner-related information, used to characterize the learner's current learning conditions and page adaptation requirements.

[0124] Learner parameters are represented as follows:

[0125] ,

[0126] in, This indicates the learner's prior knowledge level, used to assess the learner's existing foundation in the relevant knowledge; among which It indicates the learner's comprehension level and is used to determine the appropriate content presentation density and explanation pace for them; Indicates the level of interaction required, used to determine the number and complexity of interactive controls on the page; This indicates the required level of guidance, used to determine the density of prompts, feedback, and step-by-step guidance.

[0127] In this embodiment, the system performs missing parameter completion and standardization on the learner parameters. If If missing, the default setting is primary; if If missing, the default setting is medium; if If missing, the default value is 0.5; if If missing, the default setting is medium guidance. Through the above completion process, even if the user input is incomplete, the system can still obtain a computable set of learner parameters, thus ensuring that the subsequent cognitive state network construction, path optimization, and page generation processes can be executed continuously.

[0128] The technical function of this step is to: input data packets. It serves not only as the basis for generating instructional texts but also as the basis for constructing cognitive state transition networks. Learner parameters By transforming learner differences into structured variables that can be read by subsequent algorithms, interactive teaching pages can be designed to determine different segment counts, content density, interaction complexity, and feedback methods when the same teaching topic is presented to different learners.

[0129] Calling the large language model to the teaching topic and teaching materials Semantic parsing and structured extraction are performed. Based on the semantic parsing results of the teaching topic, the knowledge scope covered by the interactive teaching page, the knowledge units to be generated, and the candidate cognitive levels and target cognitive levels corresponding to each knowledge unit are determined. Based on the semantic parsing results of the teaching materials, teaching content resources associated with each knowledge unit are extracted or generated, and the prerequisite relationships between knowledge units and the executable teaching actions corresponding to each knowledge unit are extracted.

[0130] Based on the candidate cognitive levels corresponding to each knowledge unit, the content complexity of the teaching content resources, and the learner's acceptance level, the state layer target load of each knowledge unit under the corresponding candidate cognitive level is determined.

[0131] To elaborate further, for the input teaching material, the system performs corresponding structured parsing based on the material type. For text materials, it obtains the text content, conceptual keywords, and teaching objectives through title recognition, paragraph segmentation, word segmentation, keyword extraction, and recognition of teaching objective verbs. For image materials, it extracts the text from the image through optical character recognition and obtains the objects, relationships, or processes of change represented by the image through image semantic recognition. For formula materials, it converts the formula into LaTeX, MathML, or equivalent structured expressions through formula recognition and extracts the variables, operational relationships, and constraints from the formula. For code materials, it generates an abstract syntax tree through the syntax parser of the corresponding programming language and extracts functions, variables, control structures, and data dependencies.

[0132] The system uniformly converts the above parsing results into semantic fragments containing source location, text description, conceptual keywords, teaching objective verbs, and content type. Using the title and chapter boundaries as initial division boundaries, the semantic similarity between adjacent semantic fragments is calculated. When adjacent semantic fragments have the same conceptual keywords or teaching objectives, and the semantic similarity is greater than a preset threshold, they are merged. When a new core concept appears, the teaching objective changes, or the semantic similarity falls below the preset threshold, a new knowledge unit is established. This process divides the teaching material (Material) into knowledge units. Each knowledge unit At least the relevant teaching content, conceptual keywords, teaching objectives, and source location should be linked.

[0133] Pre-construct a set of labeled samples representing cognitive levels from different disciplines, with each sample including a knowledge unit. Teaching objectives, teaching actions, and corresponding cognitive level labels .in, These represent memory, understanding, application, analysis, evaluation, and creation, respectively. The labeled samples are used to train a six-class classification model, which outputs knowledge units. The probability of belonging to each cognitive level is determined, and the cognitive level is determined according to the following formula:

[0134] ,

[0135] During training and inference, the teaching actions of "memorizing, understanding, applying, analyzing, evaluating, and creating" in the teaching objectives are used as classification features. If the highest classification probability is lower than the preset confidence threshold, the large language model is called to make a secondary judgment according to the above six cognitive levels, and the low confidence results are added to the verification sample set for subsequent updates to the classification model.

[0136] Sure and Then, the system generates corresponding enhanced page section skeletons based on the teaching materials. ', and extract the segment parameter vector:

[0137] ,

[0138] in, For the number of information elements, For content density, For interaction complexity, For feedback density.

[0139] This indicates the number of information elements that can be presented independently. Each text block, image, formula, code block, example, or question is counted as one information element, therefore:

[0140] ,

[0141] This indicates content density. The system uses a page layout engine to estimate the display area occupied by each content object based on the number of characters and font size, the preset display size of images, and the layout size of formulas and code, and calculates it using the following formula:

[0142] ,

[0143] in, This indicates the estimated display area occupied by all content objects within the section. This indicates the available display area of ​​this page segment.

[0144] This represents the interaction complexity, determined by the number of controls, minimum number of operation steps, and number of conditional branches required to complete the teaching task for that section. Specifically, it is 0 for no interaction; 1 for a single click, selection, or input operation; 2 for multiple similar controls or two to three consecutive operation steps; and 3 for multiple linked controls, more than three consecutive operation steps, or conditional branches.

[0145] The feedback density is determined by the number of independent feedback messages configured in the feedback slot and the number of feedback trigger points.

[0146] ,

[0147] in, This indicates the number of independent feedback messages, such as result prompts, error explanations, correction suggestions, and step-by-step instructions. This indicates the number of interactive nodes that can trigger feedback.

[0148] The system adopts the corresponding disciplines Cognitive load of the weighted segment:

[0149] ,

[0150] Will Normalized to the load range in the training samples of this discipline And in combination with learners' acceptance level After correction, the target load of the state layer is obtained. This generates cognitive state nodes:

[0151] ,

[0152] This enables the automatic determination of cognitive levels and cognitive load for different forms of teaching materials.

[0153] The learner's mastery of each knowledge unit is determined based on at least one of the learner's historical responses, historical interaction behaviors, dwell time, number of repeated operations, and feedback results. When the historical learning data is missing, the learner's initial mastery of each knowledge unit is determined according to a preset mastery mapping rule based on the learner's prior level, which has been pre-input by the learner and processed by missing data completion and standardization.

[0154] Based on the aforementioned data packets, this embodiment models the generation process of interactive teaching pages as a Cognitive State Transition Network (CSTN). In this network, the teaching process is no longer viewed as a simple arrangement of several static page segments, but rather as a sequence of cognitive states formed by the sequential transitions of multiple cognitive state nodes:

[0155] Each state node This is a structured data object built by the system before page generation, used to represent the cognitive state corresponding to a certain stage in the teaching process. This state node not only describes the current teaching segment of the learner, but also serves as the basic processing unit for subsequent path search, segment mapping, parameter adjustment, content generation, and interactive linkage configuration.

[0156] In this embodiment, the cognitive state node is defined as:

[0157] ,

[0158] Specifically, This variable represents the cognitive level, used to characterize the learning objective level corresponding to the current state node. Its value range is the index of Bloom's hierarchy of cognitive levels {1, 2, 3, 4, 5, 6}, corresponding to memory, comprehension, application, analysis, evaluation, and creation, respectively. This variable is used to transform abstract learning objectives into computable hierarchical labels, enabling the system to determine whether a particular teaching segment should undertake teaching functions such as context introduction, phenomenon observation, variable manipulation, difference comparison, result verification, or rule induction.

[0159] This represents the target load at the state level, and its value range is... This variable describes the target cognitive load corresponding to a state node during the path planning phase, i.e., the information processing intensity that the system expects learners to experience in the teaching segment corresponding to that node. It is set by... The system can explicitly consider the load changes between adjacent teaching sections when planning the page section order, avoiding the generation of page structures with excessively large load differences between sections.

[0160] Indicates the degree of mastery, and its value range is: This variable is used to characterize the proportion of knowledge a learner has mastered in the current knowledge point or current teaching task. This can be calculated from the learner's historical responses, historical interactions, dwell time, number of repeated operations, or feedback results. In one implementation, when... At lower levels, the system may tend to generate more observation, cue, or feedback segments; when At higher levels, the system can increase the proportion of operation, comparison, or migration tasks.

[0161] Candidate cognitive state nodes are generated based on the combination of each knowledge unit, candidate cognitive level, state layer target load, and mastery level, forming a set of cognitive state nodes:

[0162] ,

[0163] The instructional actions that can be executed when transitioning between cognitive state nodes are used as state transition edges. For cognitive state nodes... and cognitive state nodes The corresponding state transition edge is represented as:

[0164] ,

[0165] in, This indicates that the learner has moved from a cognitive state node. Transfer to cognitive state node The teaching actions performed at the time include, but are not limited to, observation, operation, comparison, summarization, verification, prompting and feedback.

[0166] When cognitive state node To cognitive state node When the transition between two cognitive state nodes conforms to the teaching sequence, the prerequisite dependencies between knowledge units, the preset cognitive level transition rules, and the state level target load control requirements, a state transition edge is established between the two cognitive state nodes. The set of state transition edges is formed by all state transition edges that satisfy the above conditions:

[0167] ,

[0168] The cognitive state transition network is represented as:

[0169] ,

[0170] in, For a set of state nodes, Let be the set of state transition edges. Each node in the process corresponds to a candidate cognitive state. Each edge in the diagram corresponds to a feasible teaching action for transitioning from one cognitive state to another. These teaching actions include, but are not limited to, observation, manipulation, comparison, induction, verification, prompting, and feedback.

[0171] Based on the learner's prior knowledge level, the learner's acceptance level, and the learner's mastery of the initial knowledge units, the initial cognitive state nodes are determined from the cognitive state node set V. Based on the target knowledge units, target cognitive levels, and pre-set target mastery levels corresponding to the teaching theme, from the set of cognitive state nodes... Determine the target cognitive state node .

[0172] In the cognitive state transition network, a candidate teaching path from the initial cognitive state node to the target cognitive state node is represented as:

[0173] ,

[0174] Right now:

[0175] ,

[0176] In the state transition mechanism, if there exists a slave state node To the state node A transition edge indicates that the system can use a certain teaching action to enable the learner to move from a certain position to a certain position. The cognitive state represented enters The system does not arbitrarily arrange page segments, but first determines whether the state transition conforms to the teaching sequence, prerequisite knowledge dependencies, and load control requirements, and then selects an appropriate path based on the edge weights.

[0177] Therefore, each cognitive state node in the optimal teaching path corresponds to a teaching page segment, and the final page segment sequence is essentially a visualization and interactive presentation of the optimal cognitive state sequence on the interactive teaching page.

[0178] Through the aforementioned cognitive state transition network, this embodiment transforms the "segment arrangement problem" in traditional page generation into a "path planning problem between state nodes." This modeling approach eliminates the reliance on fixed templates or step-by-step prompts for page structure generation. Instead, it is determined by knowledge units, learner cognitive states, teaching objectives, and workload constraints, thus providing a data foundation for subsequent cognitive transition cost calculation and optimal teaching path solution.

[0179] S2. By quantifying the cognitive jump difficulty between adjacent cognitive state nodes, the cognitive transition cost is obtained. Based on the principle of minimizing cognitive transition cost, the optimal teaching path from the initial cognitive state to the target cognitive state is determined.

[0180] In this embodiment, the edge set Each edge in the algorithm represents an executable instructional action, such as observation, manipulation, comparison, induction, verification, or feedback, to transition from one cognitive state node to another. The system calculates the cognitive transition cost of the corresponding edge based on the changes in cognitive hierarchy between adjacent state nodes, changes in target cognitive load, and whether there are large leaps, and uses this cost as the edge weight.

[0181] After the edge weights are determined, and the weights of the transition edges for each state are determined, the system uses Dijkstra's algorithm, A* algorithm, or dynamic programming algorithm to solve for the optimal teaching path from the initial state to the target state from the set of candidate teaching paths that satisfy the teaching constraints.

[0182] ,

[0183] in, This represents a sequence of state nodes that has a low overall cognitive transition cost and satisfies the teaching sequence constraints. This path is not simply the one with the fewest nodes, but rather a comprehensive trade-off between cognitive smoothness, target load stability, path length, and teaching logic constraints.

[0184] The technical function of this step is that the system, through... The problem of "how to arrange page segments" is transformed into a "weighted path search problem in cognitive state space." In obtaining... Subsequently, the page generation stage directly follows this path, thus avoiding the problem of generating pages being temporarily pieced together based on local prompts, and reducing the problem of excessive cognitive spans or breaks in teaching logic between adjacent sections.

[0185] It can be explained that, for example Figure 2 As shown, this embodiment introduces the concept of Cognitive Transition Cost (CTC) to quantify the difficulty of cognitive jumps between adjacent teaching segments. Cognitive Transition Cost is used to inform the cognitive state transition network. Each state transition edge in the algorithm is assigned a computable weight, thereby transforming the page structure optimization problem into a weighted path search problem.

[0186] The cognitive transition cost function is defined as:

[0187] ,

[0188] in: Indicates the magnitude of leaps in cognitive levels. This represents the change in the target load at the state level. This represents an indicative function, indicating when a cognitive level jump exceeds a threshold. Punishment should be imposed at the appropriate time. All represent weighting coefficients, which respectively characterize the relative importance of hierarchical jump penalties, load fluctuation penalties, and large jump penalties. This represents the threshold for jumping between cognitive levels (usually set to 2, meaning that jumping more than two levels is not allowed).

[0189] In the above formula, This represents the cognitive hierarchy span between adjacent state nodes. When... A larger value indicates that learners need to make significant transitions in learning objectives between adjacent learning segments, such as moving directly from the comprehension level to the creation level. This could easily lead to learners entering higher-order tasks before they have developed sufficient cognitive preparation. Therefore, this embodiment adopts... Nonlinear terms penalize hierarchical jumps, making the cost of large cognitive jumps more significant, thus prioritizing the avoidance of cognitive cliffs when choosing a path.

[0190] This represents the change in target cognitive load between adjacent state nodes. If A larger value indicates a greater variation in the demands on the learner's working memory and information processing capabilities between adjacent page segments, potentially leading to sudden increases or decreases in workload when switching pages or progressing through segments. By... By incorporating cognitive transition costs, the system can explicitly control the load smoothness between page segments during the path planning phase.

[0191] Used to identify values ​​exceeding a preset threshold. A significant leap in cognition. When Greater than At that time, the system believes that this state transition may lead to a significant cognitive break, and then... Additional penalties are imposed on this transition. The technical role of this is to transform the teaching experience rule of "avoiding excessive jumps across levels" into a boundary weight rule that can be processed by path search algorithms.

[0192] Based on the aforementioned cognitive transition cost, the system uses a path optimization algorithm to solve the problem of transitioning from the initial state. To the target state Optimal teaching path The costs considered in path optimization include the following three categories.

[0193] First, the total transition cost:

[0194] ,

[0195] This objective aims to reduce the total number of cognitive jumps between adjacent states in the entire learning path. During path optimization, this item is made as small as possible, thereby making the cognitive progression between page segments smoother.

[0196] Second, fluctuations in cognitive load:

[0197] ,

[0198] This term is used to measure the degree of fluctuation of the state layer target load of each cognitive state node in the candidate teaching path. During the path optimization process, this term should be minimized to avoid drastic changes in information processing intensity between consecutive page segments.

[0199] Third, path steps This objective aims to control the length of the main page path, preventing the system from generating too many segments to excessively reduce the cost of cognitive transitions, which could lead to lengthy pages, decreased learning efficiency, or increased interaction burden.

[0200] This item represents the number of steps in the candidate teaching path. During path optimization, this item should be minimized to control the length of the main page path and prevent the system from generating too many page segments to excessively reduce the cost of cognitive transition, which would lead to lengthy pages, decreased learning efficiency, or increased interaction burden.

[0201] In its implementation, the system weights and sums the three types of costs according to preset weights to form candidate teaching paths. The corresponding comprehensive path cost function:

[0202] ,

[0203] in, These are used to adjust the influence of total transition cost, state-level target load fluctuation, and path steps on path selection, respectively: when emphasizing cognitive smoothness, they can improve... or When emphasizing page compactness, it can improve... .

[0204] The system determines the optimal teaching path from the set of candidate teaching paths that satisfy the teaching constraints according to the following formula:

[0205] ,

[0206] in, This represents a set of candidate teaching paths that satisfy the following constraints: knowledge unit prerequisite dependency constraint, cognitive level allowable transfer constraint, state layer target load threshold constraint, reachability constraint from initial cognitive state node to target cognitive state node, and maximum number of steps on the main page path constraint.

[0207] In one implementation, when using the A* algorithm or Dijkstra's algorithm for path solving, the system uses the cumulative cognitive transition cost, the number of path steps, the number of path nodes used to calculate the target load variance of the state layer, the cumulative load value, and the cumulative load square value as the extended search state, so that the comprehensive path cost can be calculated incrementally as the path expands; when using the dynamic programming algorithm, the system calculates the comprehensive path cost based on the above-mentioned cumulative values ​​corresponding to each candidate path, and selects the teaching path with the minimum comprehensive path cost.

[0208] By adjusting The system balances cognitive smoothness, path length, and state layer target load stability to avoid relying solely on local prompts or fixed templates to generate page segments.

[0209] S3. Determine the sequence of teaching segments based on the optimal teaching path. Each teaching segment corresponds to a cognitive state node. Based on the mapping system from educational cognition theory to page structure, map each teaching segment to a page container. Generate response content for different page containers, establish segment linkage relationships, and output the teaching page.

[0210] After determining the optimal teaching path (i.e., the cognitive state sequence), this embodiment further establishes a generation and instantiation system from cognitive state nodes to teaching page segments, which is used to transform the cognitive state sequence into an executable interactive teaching page structure.

[0211] Let the optimal teaching path be:

[0212] ,

[0213] Each state node:

[0214] ,

[0215] This represents an ordered set of state nodes from the initial state to the target state. to The order determines the arrangement order of the containers in the subsequent teaching sections. Each All are retained , , and Therefore, the path is not a regular list of pages, but a teaching progression sequence with constraints on knowledge units, cognitive levels, workload targets, and mastery levels.

[0216] This embodiment does not re-plan the segment sequence, but instead maps each state node point-by-point based on the optimal path to generate the corresponding page segment. Specifically, as follows... Figure 3 As shown, the following steps can be taken:

[0217] S31, the mapping function from cognitive level to teaching segment type.

[0218] For each state node in the path First, based on their cognitive level Determine the basic segment type and define the type mapping function:

[0219] ,

[0220] in, This represents the mapping function for segment types. Since the teaching function of a segment is primarily determined by the cognitive level, therefore... In implementation, it can depend only on explicit dependencies .

[0221] This yields the basic segment type sequence:

[0222] ,

[0223] Its function is to transform cognitive levels into segment types with clear teaching functions. The basic segment type sequence... with the optimal path In this step, the system does not readjust the segment order but generates the basic page segment skeleton according to the order of cognitive state nodes.

[0224] Specifically, based on the basic segment type and their corresponding cognitive state nodes As input, generate the basic page section skeleton. In this embodiment, the basic page segment skeleton This corresponds to the UI segment object to be instantiated. The basic page segment skeleton includes a segment type identifier, a binding relationship with the cognitive state node, layout rules, and content slots, interaction slots, and feedback slots for carrying teaching content, interactive controls, and feedback information.

[0225] Subsequently, based on the knowledge type, segment parameter vector, and teaching content resources, content is filled in, controls are selected, and the layout is determined. The basic page segment skeleton is then instantiated into a renderable, interactive graphical user interface segment. Specifically, the large language model generates the segment based on input parameters or requirements.

[0226] This ensures that the page structure originates from an optimized cognitive path, rather than a static arrangement of a preset template. For example, the memory hierarchy maps to the context introduction section, the understanding hierarchy maps to the phenomenon observation section, the application hierarchy maps to the variable manipulation section, the analysis hierarchy maps to the difference comparison section, the evaluation hierarchy maps to the result verification section, and the creation hierarchy maps to the rule summarization section.

[0227] The mapping process can be visualized as: cognitive state nodes →Basic Section Type →Basic Page Section Skeleton →Enhance page section skeleton .

[0228] Among them, the basic section type is a type label, such as: observation section, operation section; the basic section type sequence is a label sequence arranged by path, such as: introduction—observation—operation—induction; the basic page section skeleton is a structural object that can be further parameterized and instantiated, such as a section containing teaching functions, knowledge units and cognitive constraints.

[0229] S32. Correction mechanism for segment mapping based on knowledge type.

[0230] In the basic mapping Based on this, knowledge types are introduced. To modify the segment type, define the modification function:

[0231] ,

[0232] in, , representing declarative, procedural, and conditional knowledge, respectively; This represents a function to enhance the skeleton of a page section, used to modify the basic page section skeleton based on the knowledge type. Enhance at least one of the internal content structure, interaction structure, and feedback structure to obtain an enhanced page segment skeleton. For example: when and When it is an observation section, it can be enhanced to include an observation section with operation controls; when and When analyzing a segment, it can be enhanced to include a comparison segment that includes multiple instance comparisons; when Simultaneously, the difference comparison section and variable operation section are enhanced, introducing contextual change and conditional judgment structures. This correction mechanism does not change the order or number of sections, does not add new main path nodes, and only enhances the internal functional structure of the sections.

[0233] Used to adjust the knowledge type while keeping the main path unchanged. The internal structure of each section is enhanced. For declarative knowledge, enhancements may focus on definition explanations, concept clarification, or diagrammatic illustrations; for procedural knowledge, enhancements may focus on step prompts, variable manipulation controls, or operational feedback; for conditional knowledge, enhancements may focus on scenario switching, conditional judgments, multi-instance comparisons, or decision feedback. This modification only changes the content nodes, control nodes, or feedback nodes within the section, and does not change the underlying structure. The order and number of main paths are determined.

[0234] S33. Segment parameterization based on cognitive load and learner parameters.

[0235] Obtain the enhanced page section skeleton Next, structural parameters are assigned to each enhanced page segment skeleton, and segment parameter vectors are defined:

[0236] ,

[0237] in, It is the first A set of structural parameters for an enhanced page section skeleton; Enhance the set of structural parameters for page section skeletons; Control content density, Controlling interaction complexity Control the feedback density.

[0238] The system adjusts It allows for changes to the page presentation of the same teaching topic without altering the section's functional type. For example, it can reduce the number of elements and operation steps for learners with weaker foundations, while adding comparison objects or exploration tasks for learners with stronger foundations.

[0239] in, Indicates the number of information elements within a single segment. Indicates content density, Indicates the interaction complexity. Feedback density This indicates the number of segments in the main page path; the two terms have different meanings and should not be used interchangeably. Once the path is determined, it must satisfy... That is, the number of main path segments on the page corresponds to the number of state nodes in the optimal teaching path.

[0240] The set of learner parameters defined in step S1 is as follows:

[0241] ,

[0242] in, Indicates the learner's prior level. Indicates the learner's level of comprehension. Indicates the degree of interaction required. This indicates the required level of guidance.

[0243] Segment parameter vector Determined by the following function:

[0244] ,

[0245] in, This represents the function for solving segment parameters. This function is based on the corrected segment skeleton. State-level target cognitive load and learner parameters Output segment parameter vector Its technical function lies in translating learner differences into page structure differences, and influencing the actual cognitive load of each segment. Try to get as close as possible to the target cognitive load set in the path planning phase.

[0246] For example, when lower or At higher levels, the system can reduce and improve To reduce information overload and increase guided feedback; when higher or When the system is more robust, it can increase the degree of freedom in variable manipulation, the number of comparison objects, or the proportion of exploratory tasks. These adjustments do not change the segment order, but only alter the content density, interaction complexity, and feedback methods within each segment.

[0247] like If the threshold is exceeded, the system adjusts according to a preset rollback strategy. The rollback strategy may include: prioritizing reduction... Secondly, reduce Next, selectable auxiliary sections are trimmed. For those... The system does not trim the required sections of the main path that directly correspond to the teaching path, in order to ensure the integrity of the teaching path.

[0248] In one implementation, the rules governing the adjustment of segment parameters by the learner parameters include:

[0249] ,

[0250] ,

[0251] ,

[0252] ,

[0253] in, Indicates the density of basic content. Indicates the complexity of basic interactions. Indicates the basic feedback density. Indicates the number of basic segments; and These represent the scaling adjustment functions for the corresponding parameters, and their range or output results are determined by system preset rules, teacher configuration, or historical teaching data.

[0254] The technical implication of the above parameter calculation process is that the system does not simply generate the same page for all learners, but rather selects pages based on the learners' comprehension levels. Adjust content density according to interaction needs. Adjust the interaction complexity according to the guidance intensity requirements. Adjust the feedback density and based on the learner's prior knowledge. Estimate the appropriate number of segments. For learners with weaker foundations or higher guidance needs, the system tends to reduce the amount of information per screen and the complexity of interaction, while increasing feedback and prompts; for learners with better foundations or higher exploration needs, the system can increase the degree of freedom in variable manipulation, the number of comparison objects, and the proportion of exploratory tasks.

[0255] The system further calculates the actual cognitive load of a single segment:

[0256] ,

[0257] in, , , and These represent the weights of the impact of the number of information elements, interaction complexity, and content density on the cognitive load of a segment, respectively.

[0258] Regarding the weighting coefficients, it's worth explaining here. The system pre-establishes a correspondence between subject types and weighting coefficients. After teaching materials are input, the system determines the subject type based on the subject tags of the teaching materials, the course catalog, or the semantic classification results of the teaching theme. It also calls the corresponding weight coefficient and cognitive load threshold for that subject.

[0259] The weighting coefficients for different disciplines are represented by a piecewise function as follows:

[0260] ,

[0261] Among them, literature, history and language This refers to disciplines such as language, literature, history, and others that primarily rely on text comprehension and content expression; mathematics and science. Represents mathematics, physics, chemistry, and other disciplines that primarily rely on formulas, calculations, and logical derivations; programming experiments. This refers to disciplines that primarily involve interactive operations and feedback, such as programming, experimentation, and hands-on training; others. This refers to other disciplines that do not have dedicated weighting coefficients configured.

[0262] Therefore, the first The actual cognitive load of each enhanced page section skeleton is calculated segment by subject type as follows:

[0263] ,

[0264] in, Indicates the number of information elements. Indicates the interaction complexity. Indicates content density, This represents the feedback density. All parameters are normalized according to the preset value range for the corresponding discipline before calculation to avoid the influence of different parameter dimensions on the cognitive load calculation results.

[0265] The system also pre-sets cognitive load thresholds for different subjects. :

[0266] ,

[0267] The system will calculate With the corresponding disciplines Comparison. When At that time, the system reduces the interaction complexity sequentially according to the preset rollback strategy. Content density and the number of information elements and recalculate until satisfied: The aforementioned weighting coefficients and load thresholds are pre-stored in the subject parameter configuration table, eliminating the need for model training during system operation. For teaching materials where the subject type cannot be determined or where dedicated parameters have not yet been configured, the system employs other methods. The corresponding general weighting coefficients and general load thresholds.

[0268] This formula is used to convert the structural parameters within a page into comparable cognitive load values.

[0269] Full-page cognitive load is represented as:

[0270] ,

[0271] in, This represents the total cognitive load of the page, comprised of all main path segments and their necessary content. The system, based on... Determine if the overall page load exceeds a preset threshold. When the cumulative load threshold of a preset page is exceeded, the system readjusts the parameters of the corresponding section according to a preset rollback strategy. The rollback strategy includes: prioritizing the reduction of interaction complexity. Secondly, reduce content density. Secondly, reduce optional information elements or supplementary teaching content, and recalculate the actual cognitive load of the segment.

[0272] The rollback adjustment does not delete the optimal teaching path. The cognitive state nodes in the program directly correspond to the main path page segments to ensure the integrity of the teaching path and the continuity of cognitive progress.

[0273] S34, Section Instantiation and Component Generation Mechanism.

[0274] Get segment type and segment parameter vector Then, the enhanced page segment skeleton is instantiated to obtain the specific page segment:

[0275] ,

[0276] in, Indicates the first A renderable, interactive page section; This represents the page segment instantiation function, which is used to transform the enhanced page segment skeleton and segment parameter vector into the actual page segment.

[0277] Enhance page section skeleton Used to determine the teaching functions, teaching content resources, internal content structure, interaction structure, feedback structure, and layout rules of this page segment; segment parameter vector Used to determine the number of information elements, content density, interaction complexity, and feedback density within a section of the page.

[0278] like Figure 4 As shown, the instantiation function is performed through the page section. The system outputs no longer plain text explanations, but rather executable page segment objects containing teaching content, interactive controls, feedback components, and layout structure. Specifically, The execution process includes:

[0279] First, based on the basic segment type and enhance the page section skeleton Define the teaching function roles of each section. For example, the introduction section is used to generate problem situations and task objectives; the observation section is used to generate diagrams, examples, or dynamic phenomena; the operation section is used to generate variable controls; the comparison section is used to generate content comparing multiple objects or conditions; the summarization section is used to generate pattern hints or conclusions; and the verification section is used to generate questions, judgment tasks, and feedback content.

[0280] Second, based on cognitive state nodes Determine the knowledge unit, cognitive level, target load, and mastery level constraints corresponding to this segment. If If the level is low, the system can add prompts, examples, or immediate feedback in that section; if With a higher score, the system can reduce repetitive explanations and increase exploration tasks.

[0281] Third, according to Determine the number of information elements, content density, interaction complexity, and feedback density within each segment. For example, Control the number of examples or illustrations. Control the degree of compression of text and images. Control the number of controls and the operation steps. Frequency of control prompts and feedback.

[0282] Fourth, the system automatically selects interactive controls based on the data type of the interactive variables in the teaching content resources. Continuous variables correspond to slider controls, discrete variables to selection controls, text input to input box components, and question answering to question components and submission components. This mechanism ensures that teaching variables are presented on the page in a way that suits their data type.

[0283] Fifth, the layout is generated based on the number, type, and relationship structure of objects within a section. When a section contains two main objects that need to be compared, the system uses a left-right layout; when it contains multiple objects or multiple conditions, the system uses a grid layout; when the content has a process sequence, the system can use a step list, flowchart, or timeline layout.

[0284] Through the above segment instantiation mechanism, cognitive state nodes are transformed into page segment objects that can be displayed, manipulated, and fed back on in terminal devices, thereby realizing the generation of cognitive path into page structure.

[0285] Based on the order of cognitive state nodes in the optimal teaching path, the instantiated page segments are... Arrange them to form a sequence of teaching sections:

[0286] ,

[0287] S35, Inter-segment linkage generation mechanism.

[0288] like Figure 6 As shown, in generating each page segment Subsequently, the system further establishes the interactive response relationship between sections. The linkage function is defined as:

[0289] ,

[0290] in, Indicates the interactive trigger section. Indicates the target segment that was triggered for updating. This indicates the interactive and interconnected relationship between the two.

[0291] The linkage relationship describes how a learner's actions in one page segment trigger state updates for target objects in other page segments. The update rules include deterministic update rules and generative update rules.

[0292] Deterministic update rules are used to calculate the update result of the target object based on the parameters carried by the interaction event, through preset calculation functions, data binding relationships, or component state transition rules. For example, they can be used to recalculate function image data based on variable parameters, update comparison results, change the state of interactive controls, or display preset pattern prompts.

[0293] Generative update rules are used to invoke a large language model to generate corresponding update content based on interaction events, learner status, and teaching context when explanatory text, personalized prompts, or open-ended feedback are needed.

[0294] After the target object completes its state update, the page rendering engine performs partial re-rendering on the corresponding content objects, image objects, interactive controls, or feedback components in the target page segment based on the updated object state.

[0295] For example, when learners adjust variables in the variable manipulation section, the system can trigger the phenomenon observation section to update images, examples, or dynamic phenomena, trigger the difference comparison section to update the comparison results, and trigger the rule summarization section to display or update rule prompts when preset conditions are met; when learners submit their answers in the result verification section, the system can trigger the feedback component to output explanations, prompts, or correction suggestions.

[0296] When the same triggering event corresponds to multiple target page segments, the system executes the updates of the target page segments sequentially according to a preset linkage priority. In one implementation, for page segment updates triggered by variable operation events, the linkage priority from high to low is: observation > comparison > induction > verification.

[0297] The system prioritizes updating observation results, then comparison results, followed by pattern summaries and prompts, and finally verification or feedback information. By performing state calculations and partial re-rendering according to this priority, data dependency conflicts or display order disorder caused by simultaneous updates of multiple page segments can be reduced.

[0298] In one implementation, the segment linkage relationship includes: the operation segment triggers the observation segment update, the operation segment triggers the difference comparison segment update, the operation segment triggers the conclusion display in the rule summarization segment, and the verification segment triggers the feedback area update. Specifically, when the learner adjusts parameters in the variable operation segment, the system first updates the image, example, or dynamic phenomenon in the observation segment according to the parameters; then, it updates the comparison results in the comparison segment according to the updated observation results or the parameters; when the parameter change meets preset conditions, it updates the rule prompts in the summarization segment; after the learner submits the answer in the verification segment, the system uses deterministic or generative update rules to cause the feedback component to output explanatory feedback, prompts, or correction suggestions based on the answer.

[0299] When the verification result does not meet the preset conditions, the feedback component can also trigger the update of prompt information, control status or operation conditions in the variable operation section to guide the learner to re-execute the corresponding operation.

[0300] The technical advantage of this step is that the generated teaching page is no longer a static stack of content, but rather possesses the ability to transmit information, update status, and respond interactively across page segments. A learner's actions within one page segment can drive changes in content objects, image objects, interactive controls, or feedback components in other page segments, thus forming a continuous, executable, and responsive interactive teaching page.

[0301] S36. Output the teaching page.

[0302] Output the final teaching page. The output includes the page structure, teaching segment sequence, segment content, interactive controls, feedback components, layout structure, interactive linkage rules, and optional streaming incremental output data.

[0303] The page structure defines the arrangement of sections on the page; the section content carries teaching explanations, examples, tasks, and feedback; the interactive controls receive learner actions; the feedback components output prompts, explanations, or correction suggestions; the layout structure controls the page presentation; and the interactive linkage rules define the cross-section response logic after a learner's action.

[0304] In one implementation, the system can follow the optimal teaching path. The corresponding segment sequence progressively outputs the page skeleton, segment content, interactive controls, and linkage rules, allowing the terminal device to display the generated segments before the complete page is generated, and supporting learners to interact with the generated segments. Therefore, this embodiment can adapt to streaming generation scenarios using large language models, multimodal generation modules, or front-end dynamic rendering frameworks.

[0305] Example 2 - Specific Application Example.

[0306] The system uses "Changes in the Graph of a Quadratic Function" as its teaching theme. It receives teaching materials including the definition of a quadratic function, examples of function graphs, explanations of parameter changes, and practice problems. Learner parameters are set as follows:

[0307] ,

[0308] in, = Elementary learners have a weaker prior knowledge of the changing patterns of quadratic function graphs; = Slow learners suggest that the system should reduce the speed of content delivery and the density of information displayed on the same screen; = 0.4 indicates that the interaction requirements are at a low to medium level, and the system should not be equipped with too many complex controls; = Strong boot means the system should provide more step prompts, error explanations, and immediate feedback.

[0309] S1. Determine the optimal teaching path and segment mapping.

[0310] The system performs semantic analysis on the teaching topics and teaching materials, and obtains the knowledge content involved in this embodiment, including the general form of a quadratic function, the relationship between parameter a and the opening direction and width of the parabola, the relationship between parameter b and the axis of symmetry and the vertex position, the relationship between parameter c and the y-intercept and the vertical translation of the graph, and the task of judging the changes in the function graph based on the changes in parameters.

[0311] Based on the aforementioned knowledge content, learner parameters, and instructional constraints, the system constructs a cognitive state transition network and determines the optimal instructional path from among the candidate instructional paths:

[0312] ,

[0313] The corresponding cognitive path is: .

[0314] in, The contextual introduction state corresponds to the task of changing the graph of a quadratic function. Observe the state of the quadratic function graph and the changes in its parameters accordingly. By matching the operation parameters and observing the changes in the image, Correspondingly analyze and summarize the state of the relationship between parameters and image changes. This corresponds to the state of verifying the summarized rules through questions. The above cognitive states correspond to the cognitive levels of memory, understanding, application, analysis, and evaluation, respectively.

[0315] Based on the pre-defined mapping relationship between cognitive levels and segment types, the system obtains the following segments: context introduction segment, phenomenon observation segment, variable manipulation segment, rule induction segment, and result verification segment. Therefore, the teaching segment order in this embodiment is as follows:

[0316] Introduction → Observation → Operation → Induction → Verification

[0317] The order of these segments is consistent with the order of cognitive state nodes in the optimal teaching path, and the system does not readjust the segment order during the page generation stage.

[0318] The system generates five basic page segment skeletons based on each basic segment type and its corresponding cognitive state node. (Segment skeletons are then introduced.) This includes learning task slots, problem context slots, learning objective slots, and guidance prompt slots; observation section skeleton. Includes slots for displaying function graphs, function expressions, parameter values, and phenomena; and a skeleton for the operation section. This includes slots for parameters a, b, and c, a slot for displaying the current parameter value, and a slot for operation prompts; it also summarizes the section skeleton. This includes slots for pattern hints, condition comparisons, learner inductive input, and conclusion display; and a validation section skeleton. This includes question slots, answer control slots, submission control slots, and feedback component slots.

[0319] The above basic page section skeleton only determines the teaching functions, internal slots, and basic layout of each section; the specific teaching content, number of interactive controls, feedback frequency, and display density have not yet been determined.

[0320] The system identifies that the teaching content in this embodiment simultaneously includes declarative knowledge, procedural knowledge, and conditional knowledge. Specifically, the general form of a quadratic function and the meaning of its parameters belong to declarative knowledge; the process of adjusting parameters and observing image changes belongs to procedural knowledge; and the patterns of image changes under different parameter conditions belong to conditional knowledge.

[0321] Based on the above knowledge types, the system enhances the basic page segment skeleton. (Observe the segment skeleton.) The code adds a structure linking function expressions and function graphs, a structure synchronously displaying current parameter values ​​and graph changes, and a structure comparing graph changes before and after; the operation segment skeleton... The document adds continuous variable control structures for parameters a, b, and c, parameter value range constraints, control change events, data binding relationships between control events and observation sections, and operation prompts; it also summarizes the section skeleton. The code adds parameter condition judgment structure, image feature comparison structure, pattern prompt display condition, and learner induction result input structure; and verifies the segment skeleton. The system adds a structure for judging the answer result, a structure for identifying the error type, a structure for explanatory feedback, and a guiding structure for returning to the operation section or observation section.

[0322] After enhancement, the enhanced page segment skeletons were obtained respectively. , , and The enhancement process does not change the number or order of the five main path page segments; it only changes the content structure, interaction structure, and feedback structure within the skeleton of each page segment.

[0323] S2. Determine the segment parameters and complete the instantiation.

[0324] The system determines the structural parameters of each segment based on learner parameters, the state layer target load of each page segment, and the enhanced page segment skeleton.

[0325] In this embodiment, the introduction section includes three main information elements: learning background, learning question, and learning objective. The content density is set to low, the interaction complexity to 0, and the feedback density to 1. The observation section includes four main information elements: function expression, current parameter value, function graph, and graph feature description. The content density is set to medium-low, the interaction complexity to 1, and the feedback density to 2.

[0326] The operation section includes four main information elements: parameter a slider, parameter b slider, parameter c slider, and operation prompts. Its interaction complexity is set as follows: .

[0327] Feedback density is set as follows: .

[0328] This setting indicates that the operation section retains only three basic parameter controls and necessary operation prompts, and does not set up complex parameter combination experiments, simultaneous operation of multiple images, or multi-step exploration tasks.

[0329] The summary section includes three main information elements: condition prompts, a pattern completion box, and a pattern conclusion. The interaction complexity is set to 1, and the feedback density is set to: .

[0330] The verification section includes three main information elements: verification questions, answer controls, and feedback components. The interaction complexity is set to 1, and the feedback density is set to: 2.

[0331] The above parameter settings reflect the learner's characteristics of being a beginner, slow learner, and highly guided learner. That is, the system limits the amount of information displayed on the screen and complex operations, while increasing the frequency of prompts and feedback.

[0332] Following the method for calculating actual cognitive load for each segment described above, the system obtained the actual cognitive load for the five page segments as follows: .

[0333] The cumulative cognitive load of the page is: .

[0334] In this embodiment, the cumulative cognitive load threshold for the page, determined based on the learner parameters, is 6.0. Since the cumulative cognitive load for the page does not exceed the threshold, and the actual cognitive load of each page segment is within the tolerance range of the target load of the corresponding state layer, there is no need to delete the main path page segment, nor is it necessary to re-execute path planning or parameter rollback adjustment.

[0335] Subsequently, the system writes the specific teaching content, interactive controls, feedback components, and layout rules into the corresponding enhanced page segment skeleton, generating five renderable and interactive page segments.

[0336] Introducing page sections The question is written in the middle: "In quadratic functions..." In the example, how does the graph of the function change when parameters a, b, and c change? The learning objectives include observing parameter changes, manipulating parameters, summarizing patterns, and completing verification tasks.

[0337] Observe the page section This generates a coordinate system, the graph of the quadratic function, the current function expression, the current parameter values, and a display area for the opening direction, axis of symmetry, vertex, and y-intercept. The initial function is set as follows: .

[0338] The initial parameters are set as follows: ,

[0339] Operation page section Three sliders are generated. The range of values ​​for parameter 'a' is set to... Furthermore, excluding 0, the value ranges of parameters b and c are both set to... All three sliders are bound to parameter change events. When the learner completes the parameter adjustment, the system reads the new parameter value and triggers an update of the relevant page section.

[0340] Summarize page sections The text is set to fill in content according to a pattern, including "when..." When a < 0, the parabola opens in (a certain direction). When |a| increases, the parabola opens in (a certain direction). When parameter c increases, the function graph shifts as a whole in (a certain direction). The system gradually displays pattern prompts based on the learner's parameter operation results, rather than displaying the complete conclusion all at once.

[0341] Verification page section The question generated in the middle is: "Put the function..." Become What changes occurred in the function graph afterward? (Also, option controls, submit controls, and feedback components are provided.)

[0342] S3. Establish linkage relationships and output the page.

[0343] When learners are operating the page section When parameters a, b, or c are adjusted, the system reads the changed parameter values ​​and recalculates the quadratic function graph data.

[0344] For example, when the learner sets the parameter to:

[0345] ,

[0346] Observe the page section The function expression in the file is updated to:

[0347] ,

[0348] The function graph is synchronously updated to a parabola that opens downwards, has a narrow opening, and is shifted upwards by 3 units. The synchronously updated graph features are: downward opening, narrow opening width, axis of symmetry x=0, vertex position (0,3), and y-intercept of 3.

[0349] At the same time, summarize page sections The pattern hint has been updated to: "When the parameter 'a' changes from a positive number to a negative number, the direction of the parabola's opening changes."

[0350] When learners are in the verification page section After the learner submits their answer, the system updates the feedback component based on the response. If the learner answers correctly, the feedback component displays: "Correct answer. You have been able to determine the main effects of parameters a and c on the function's graph." If the learner answers incorrectly, the system generates explanatory feedback based on the incorrect answer, such as: "Please re-observe the sign of parameter a. When parameter a changes from positive to negative, the direction of the parabola's opening changes." Simultaneously, the parameter a slider in the operation page section is highlighted to guide the learner to re-operate and observe the graph changes.

[0351] The parameter calculation, function graph update, and control state update are performed by preset calculation rules and the page execution program. When personalized error explanations are needed, the system calls the large language model to generate feedback content based on the answer, the current learner's state, and the teaching context. After completing the target object state update, the page rendering engine only performs partial re-rendering on the affected page sections.

[0352] Through the above linkage process, the page is generated as follows: A closed-loop interaction.

[0353] The final generated interactive teaching page includes an introductory page section. Observe the page section Operation page section Summarize page sections and verification page section Each page segment is arranged in the order of cognitive state nodes in the optimal teaching path, and the aforementioned segment linkage relationships are loaded.

[0354] For learners in this embodiment who are beginners, learn slowly, and have a strong need for guidance, the system controls the number of information elements on the same screen, the number of examples, and the complexity of interactive operations, while providing a high frequency of prompts and feedback. For learners with higher prior knowledge or higher interaction needs, the system can add tasks such as simultaneous comparison of multiple function graphs, parameter combination, and exploratory questions, and appropriately reduce repetitive prompts, thereby generating teaching pages with higher interactive complexity and exploratory nature.

[0355] Comparative Example 1: Page construction method that does not employ cognitive state transfer.

[0356] To verify the technical effect of the present invention, a comparative example was set up, and pages with the same teaching theme were constructed using the traditional fixed template method.

[0357] Comparison method: A pre-set fixed segment sequence is used (context introduction → phenomenon observation → variable operation → pattern induction → result verification), without considering the cost of cognitive level jump.

[0358] Comparison results:

[0359] This fixed sequence omits the "difference comparison section" and the "result validation section," resulting in errors from "variable manipulation" (application level). Jump directly to "Law Induction" (creating hierarchies, ), cognitive level leap Exceeding the threshold This creates a cognitive cliff.

[0360] Experimental data shows that learners' confusion level in the rule summarization section was significantly higher than that of the method in this invention (confusion score 4.2 vs 2.1). The completion time has been extended by 35%.

[0361] Learner feedback shows that 62% of learners reported "a sudden jump to the summary of rules with a lack of transition," while only 12% of learners reported similar feedback regarding the method of this invention.

[0362] Conclusion: Comparative analysis shows that page construction methods that do not employ cognitive state transition and transition cost optimization are prone to cognitive cliffs, reducing learning effectiveness.

[0363] Comparative Example 2: Page building methods that do not employ the optimal path algorithm.

[0364] Comparison method: A greedy algorithm is used, which selects the next segment that is locally optimal each time (only minimizing the single-step load change) without considering the global transition cost.

[0365] Comparison results:

[0366] The greedy algorithm generates the following sequence: situation introduction → phenomenon observation → variable operation → result verification → pattern induction. Placing "result verification" before "pattern induction" results in an illogical logical order.

[0367] From the results verification (evaluation level, From generalizing patterns to creating hierarchical structures, Although the cognitive leap cost is low, it undermines the teaching logic of "induction first, verification later".

[0368] Actual test data shows that the knowledge retention rate of this sequence (tested after 7 days) is 58%, while the optimal path method of this invention is 78%, showing a significant difference. ).

[0369] Conclusion: Comparative analysis shows that using only a local greedy strategy without global path optimization can easily lead to logically illogical segment sequences, affecting long-term learning performance.

[0370] Based on the same idea as the cognitive state transition-based page adaptive construction method in the above embodiments, the present invention also provides a cognitive state transition-based page adaptive construction system, which can be used to execute the above-described cognitive state transition-based page adaptive construction method. For ease of explanation, the structural diagram of the embodiment of the cognitive state transition-based page adaptive construction system only shows the parts related to the embodiments of the present invention. Those skilled in the art will understand that the illustrated structure does not constitute a limitation on the device, and may include more or fewer components than illustrated, or combine certain components, or have different component arrangements.

[0371] Please see Figure 6 In another embodiment of this application, a page adaptive construction system based on cognitive state transition is provided. The system includes an input module ①, a state network construction module ②, a path optimization module ③, a segment mapping module ④, a parameter adjustment module ⑤, a content generation and writing module ⑥, a linkage establishment module ⑦, and an output module ⑧.

[0372] Input module ① is used to receive teaching topics, teaching materials and learner parameters, and form an input data packet;

[0373] Network construction module ② is used to model the generation process of interactive teaching pages as a cognitive state transition network, wherein each cognitive state node in the cognitive state transition network represents the cognitive state at a certain moment in the teaching process;

[0374] The path optimization module ③ is used to obtain the cognitive transition cost by quantifying the cognitive jump difficulty between adjacent cognitive state nodes, and to determine the optimal teaching path from the initial cognitive state to the target cognitive state based on the principle of minimizing cognitive transition cost.

[0375] Segment mapping module ④ is used to determine the sequence of teaching segments according to the optimal teaching path. Each teaching segment corresponds to a cognitive state node, based on the mapping system from educational cognition theory to page structure.

[0376] The parameter adjustment module ⑤ is used by the system to calculate the segment parameter vector for each teaching segment after determining the corrected segment type, adjust the segment parameters, and determine the segment parameter vector.

[0377] Content generation and writing module ⑥ is used to map each teaching section into a page container;

[0378] Linkage Establishment Module ⑦ is used to establish the linkage relationship between the interactive triggering section and the update target section, so as to enable learners to trigger content updates across sections when operating in a single section;

[0379] Output module ⑧ is used to output the teaching pages.

[0380] It should be noted that the page adaptive construction system based on cognitive state transition of the present invention corresponds one-to-one with the page adaptive construction method based on cognitive state transition of the present invention. The technical features and beneficial effects described in the embodiments of the page adaptive construction method based on cognitive state transition described above are all applicable to the embodiments of the page adaptive construction method based on cognitive state transition. For details, please refer to the description in the embodiments of the method of the present invention, which will not be repeated here.

[0381] Furthermore, in the implementation of the page adaptive building system based on cognitive state transition in the above embodiments, the logical division of each program module is only an example. In actual applications, the above functions can be assigned to different program modules as needed, for example, for the configuration requirements of the corresponding hardware or for the convenience of software implementation. That is, the internal structure of the page adaptive building system based on cognitive state transition is divided into different program modules to complete all or part of the functions described above.

[0382] Please see Figure 7 In one embodiment, an electronic device is provided that implements a page adaptive construction method based on cognitive state transition. The electronic device 20 may include a first processor 21, a first memory 22 and a bus, and may also include a computer program stored in the first memory 22 and executable on the first processor 21, such as a page adaptive construction program 23 based on cognitive state transition.

[0383] The first memory 22 includes at least one type of readable storage medium, including flash memory, portable hard drive, multimedia card, card-type memory (e.g., SD or DX memory), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the first memory 22 can be an internal storage unit of the electronic device 20, such as the portable hard drive of the electronic device 20. In other embodiments, the first memory 22 can be an external storage device of the electronic device 20, such as a plug-in portable hard drive, smart media card (SMC), secure digital card (SD), flash card, etc., equipped on the electronic device 20. Furthermore, the first memory 22 can include both internal storage units and external storage devices of the electronic device 20. The first memory 22 can be used not only to store application software and various types of data installed on the electronic device 20, such as the code of the cognitive state transition-based page adaptive builder 23, but also to temporarily store data that has been output or will be output.

[0384] In some embodiments, the first processor 21 may be composed of integrated circuits, such as a single packaged integrated circuit or multiple integrated circuits with the same or different functions, including combinations of one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and various control chips. The first processor 21 is the control unit of the electronic device, connecting various components of the entire electronic device through various interfaces and lines. It executes programs or modules stored in the first memory 22 and calls data stored in the first memory 22 to perform various functions of the electronic device 20 and process data.

[0385] Figure 7 Only electronic devices with components are shown; it will be understood by those skilled in the art that... Figure 7 The structure shown does not constitute a limitation on the electronic device 20, and may include fewer or more components than shown, or combine certain components, or have different component arrangements.

[0386] The page adaptive construction program 23 based on cognitive state transition stored in the first memory 22 of the electronic device 20 is a combination of multiple instructions. When run in the first processor 21, it can achieve the following:

[0387] The generation process of interactive teaching pages is modeled as a cognitive state transition network, where each cognitive state node represents a cognitive state at a certain moment in the teaching process.

[0388] The cognitive transition cost is obtained by quantifying the difficulty of cognitive jumps between adjacent cognitive state nodes. Based on the principle of minimizing cognitive transition cost, the optimal teaching path from the initial cognitive state to the target cognitive state is determined.

[0389] The teaching segment sequence is determined based on the optimal teaching path. Each teaching segment corresponds to a cognitive state node. Based on the mapping system from educational cognition theory to page structure, each teaching segment is mapped to a page container. Response content is generated for different page containers, and the segment linkage relationship is established to output the teaching page.

[0390] Furthermore, if the modules / units integrated in the electronic device 20 are implemented as software functional units and sold or used as independent products, they can be stored in a non-volatile computer-readable storage medium. The computer-readable medium may include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, or a read-only memory (ROM).

[0391] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments described above. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0392] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0393] The above embodiments are preferred embodiments of the present invention, but the embodiments of the present invention are not limited to the above embodiments. Any changes, modifications, substitutions, combinations, or simplifications made without departing from the spirit and principle of the present invention shall be considered equivalent substitutions and shall be included within the protection scope of the present invention.

Claims

1. A page adaptive construction method based on cognitive state transition, characterized in that, Includes the following steps: The generation process of interactive teaching pages is modeled as a cognitive state transition network, where each cognitive state node represents a cognitive state at a certain moment in the teaching process. The cognitive transition cost is obtained by quantifying the difficulty of cognitive jumps between adjacent cognitive state nodes. Based on the principle of minimizing cognitive transition cost, the optimal teaching path from the initial cognitive state to the target cognitive state is determined. The teaching segment sequence is determined based on the optimal teaching path. Each teaching segment corresponds to a cognitive state node. Based on the mapping system from educational cognition theory to page structure, each teaching segment is mapped to a page container. Response content is generated for different page containers, and the segment linkage relationship is established to output the teaching page.

2. The page adaptive construction method based on cognitive state transition according to claim 1, characterized in that, The process of generating interactive teaching pages is modeled as a cognitive state transition network, specifically as follows: It receives teaching topics, teaching materials, and learner parameters, and forms an input data package; Perform missing parameter completion and standardization on the learner parameters to obtain a computable set of learner parameters. The learner parameter set is represented as follows: ， in, Indicates the learner's prior level. Indicates the learner's level of comprehension. Indicates the degree of interaction required. Indicates the required intensity of guidance; The large language model is invoked to perform semantic parsing and structured extraction on the teaching topics and teaching materials. Based on the semantic parsing results of the teaching topics, the knowledge scope covered by the interactive teaching page, the knowledge units to be generated, and the target cognitive level corresponding to each knowledge unit are determined. Based on the semantic parsing results of the teaching materials, teaching content resources associated with each knowledge unit are generated, and the prerequisite relationships between knowledge units and the executable teaching actions corresponding to each knowledge unit are extracted. Cognitive state nodes are defined by combining each knowledge unit, the corresponding cognitive level, the target load of the state layer, and the learner's mastery level. Based on at least one of the learner's historical responses, historical interaction behaviors, dwell time, number of repeated operations, and feedback results for each knowledge unit, the learner's mastery level of each knowledge unit is determined; when the historical learning data is missing, the learner's prior level in the learner parameters pre-input by the learner is obtained, the learner's prior level is processed for missing information completion and standardization, and the learner's initial mastery level of each knowledge unit is determined according to the preset mastery level mapping rules; Cognitive state nodes are defined by combining knowledge units, cognitive levels, state-level target load, and mastery levels. As shown in the following formula: ， in, Represents a knowledge unit. Indicates cognitive level, Represents cognitive state nodes The corresponding state layer target load, This indicates the learner's level of mastery of the corresponding knowledge unit; The executable teaching actions corresponding to the transition between cognitive state nodes are used as state transition edges. The range of state transition edges that can be established is limited according to the prior relationships between the knowledge units, and a set of state transition edges is obtained. Based on the learner's prior knowledge level, the learner's acceptance level, and the learner's mastery of the corresponding knowledge unit, the initial cognitive state node is determined from the cognitive state nodes; based on the target knowledge unit, target cognitive level, and preset target mastery level corresponding to the teaching topic, the target cognitive state node is determined. Based on the set of cognitive state nodes and the set of state transition edges, a cognitive state transition network is constructed.

3. The page adaptive construction method based on cognitive state transition according to claim 1, characterized in that, The cognitive transition cost is obtained by quantifying the difficulty of cognitive jumps between adjacent cognitive state nodes. Specifically, the cognitive transition cost function is defined as follows: ， in, Indicates the magnitude of cognitive level jumps. This represents the change in target load at the state level. This represents an indicative function, indicating when a cognitive level jump exceeds a threshold. Punishment will be imposed at that time.

4. The page adaptive construction method based on cognitive state transition according to claim 1, characterized in that, The optimization objectives of the teaching path include minimizing the total transition cost, minimizing cognitive load fluctuations, and minimizing the number of steps in the path; The total transition cost, cognitive load fluctuation, and minimum path steps are weighted and summed according to preset weights to obtain the comprehensive path cost corresponding to the candidate teaching path. The minimized total transition cost is used to reduce the total number of cognitive jumps between adjacent cognitive state nodes in the entire teaching path; the minimized cognitive load fluctuation is used to reduce the fluctuation of the state layer target load of each cognitive state node in the teaching path; and the minimized path steps are used to control the length of the main path of the page. For candidate teaching paths: ， Its integrated path cost function is: ， The optimal teaching path is determined according to the following formula: ， in, This represents the i-th cognitive state node. Represents cognitive state nodes The corresponding state layer target load, where T represents the number of path steps in the candidate teaching path. These are used to adjust the influence of total transition cost, cognitive load fluctuation, and path steps on path selection, respectively. This represents the set of candidate teaching paths that satisfy the teaching constraints. CTC (·) represents the cognitive transition cost function. Var (·) represents the cognitive load fluctuation function; The teaching constraints include at least the prerequisite dependency constraints between knowledge units, the cognitive level transfer constraints, the state layer target load threshold constraints, the accessibility constraints, and the maximum number of steps on the main page path constraints. The prerequisite dependency constraint is used to limit the cognitive state node corresponding to the prerequisite knowledge unit to be located before the cognitive state node corresponding to the target knowledge unit when the learner's mastery of the prerequisite knowledge unit is lower than a preset mastery threshold; when the learner's mastery of the prerequisite knowledge unit reaches the preset mastery threshold, the candidate teaching path is allowed not to include the cognitive state node corresponding to the prerequisite knowledge unit. The cognitive level transfer constraint is used to determine, based on a preset cognitive level transfer matrix, whether a transfer between cognitive levels corresponding to adjacent cognitive state nodes is allowed, and satisfies the following: ， in, This represents the preset cognitive level transition matrix. and These represent adjacent cognitive state nodes. and The corresponding cognitive level, with a matrix element value of 1 indicating that transitions between the corresponding cognitive levels are allowed; The state-level target load threshold constraint is used to limit the state-level target load corresponding to each cognitive state node in the candidate teaching path from not exceeding the state-level target load threshold determined according to the learner parameters, and satisfies: ， in, Represents cognitive state nodes The corresponding state layer target load, Indicates based on learner parameters The maximum target load at the determined state level; The reachability constraint is used to limit the candidate teaching path to start from the initial cognitive state node and end at the target cognitive state node, and to ensure that there is a corresponding state transition edge and an executable teaching action between any two adjacent cognitive state nodes; the maximum number of steps in the main page path constraint is used to limit the number of steps in the candidate teaching path to not exceed a preset maximum number of steps, and to satisfy: , , in, This represents the initial cognitive state node. Represents the target cognitive state node. This represents the set of state transition edges in a cognitive state transition network. Represents cognitive state nodes Transfer to cognitive state node The corresponding executable teaching actions, This indicates the number of steps in a candidate teaching path. This indicates the maximum number of steps in the preset main page path.

5. The page adaptive construction method based on cognitive state transition according to claim 1, characterized in that, The step of determining the sequence of teaching segments based on the optimal teaching path, where each teaching segment corresponds to a cognitive state node, is based on a mapping system from educational cognition theory to page structure, specifically as follows: For each cognitive state node in the optimal teaching path Based on the cognitive level corresponding to the cognitive state node, the corresponding basic segment type is determined through a preset segment type mapping relationship. Based on the arrangement order of the cognitive state nodes in the optimal teaching path, a basic segment type sequence is formed: ， According to the aforementioned basic segment type sequence, each basic segment type and their corresponding cognitive state nodes As input, generate basic page segment skeletons that correspond one-to-one with each cognitive state node. The basic page segment skeleton is a graphical user interface segment object to be instantiated, including a segment type identifier, a binding relationship with cognitive state nodes, and content slots, interaction slots, and feedback slots for carrying teaching content, interactive controls, and feedback information; wherein, the basic segment type is used to determine the teaching function of the basic page segment skeleton, and the cognitive state nodes are used to determine the knowledge unit, cognitive level, state layer target load, and mastery level constraints corresponding to the basic page segment skeleton; Based on the knowledge type corresponding to each knowledge unit, the corresponding basic page section skeleton is... Enhance at least one of the internal content structure, interaction structure, and feedback structure to obtain an enhanced page segment skeleton. The knowledge types are identified by the large language model based on the teaching theme and teaching materials, including declarative knowledge, procedural knowledge, and conditional knowledge; wherein, the enhancement process maintains the arrangement order and number of cognitive state nodes in the optimal teaching path, and the enhanced page segment skeleton... The number corresponds to the number of cognitive state nodes in the optimal teaching path; Enhance the page section skeleton Assign structure parameters, define the first The segment parameter vector corresponding to the skeleton of each enhanced page segment: ， in, Indicates the number of information elements. Indicates content density, Indicates the interaction complexity. Indicates feedback density; The segment parameter vector is converted into a computable cognitive load value, the first... The actual cognitive load of an enhanced page section skeleton is: ， in, , and , where are weighting coefficients, representing the degree of influence of the number of information elements, interaction complexity, content density, and feedback density on the cognitive load of the segment, respectively; Define learner parameters As shown in the following formula: ， in, Indicates the learner's prior level. Indicates the learner's level of comprehension. Indicates the degree of interaction required. Indicates the required intensity of guidance; The segment structure parameters are adjusted based on the learner parameters, as shown in the following formula: ， ， ， ， in, This represents the basic content density of the i-th enhanced page segment skeleton. Indicates the complexity of basic interactions. Indicates the basic feedback density. This indicates the number of basic information elements in the skeleton of the i-th enhanced page segment. and These represent the scaling adjustment functions for the corresponding learner parameters, and their range or output results are determined by system preset rules, teacher configuration, or historical teaching data. Based on the enhanced page segment skeleton Determine the segment parameter vector As shown in the following formula: ， in, Represents the function for solving segment parameters. Represents cognitive state nodes The corresponding state layer target load and learner parameters The segment parameter vector is determined by the segment parameter solution function. ; The determined segment parameter vector Write the corresponding enhanced page section skeleton Based on the teaching content, content structure, interaction structure, feedback structure, and layout rules in the enhanced page segment skeleton, as well as the number of information elements, content density, interaction complexity, and feedback density limited by the segment parameter vector, the enhanced page segment skeleton is instantiated to obtain renderable and interactive page segments. As shown in the following formula: ， in, This function instantiates a page segment, which is used to write teaching content into the enhanced page segment skeleton, generate corresponding interactive controls and feedback components, and determine the page layout based on the objects inside the segment and their relationships.

6. The page adaptive construction method based on cognitive state transition according to claim 5, characterized in that, The mapping of each teaching segment to a page container specifically involves: Based on the enhanced page segment skeleton Determine the teaching function of each section and generate the corresponding internal objects; The internal object structure corresponding to the segment is determined based on the cognitive state node, including cognitive level, target load and mastery level constraints, and the problem situation or task background of the segment is adjusted using the mastery level constraints. Based on the segment parameter vector Determine the page variables within the section, including the number of information elements, content density, interaction complexity, and feedback density; Automatically select interactive controls based on the characteristics of page variables, including: slider controls for continuous variables, selection controls for discrete variables, input box controls for text input, and question and submission components for true / false or multiple-choice questions. The page layout is determined based on the number, type, and relationships of internal objects, and the teaching content, interactive controls, and feedback components are written into the corresponding content slots, interaction slots, and feedback slots to generate the renderable and interactive page segments. .

7. The page adaptive construction method based on cognitive state transition according to claim 5, characterized in that, The establishment of the segment linkage relationship includes: Establish a linkage between the interactive triggering section and the update target section, so that learners can trigger content updates across sections when they operate in a single section; When the same operation triggers synchronous responses in multiple segments, updates are performed according to preset linkage priorities, which are observation, comparison, summarization, and verification from high to low.

8. A page adaptive construction system based on cognitive state transition, characterized in that, The page adaptive construction method based on cognitive state transition applied to any one of claims 1-7 includes: an input module, a state network construction module, a path optimization module, a segment mapping module, a parameter adjustment module, a content generation and writing module, a linkage establishment module, and an output module; The input module is used to receive teaching topics, teaching materials, and learner parameters, and form an input data package; The network construction module is used to model the generation process of interactive teaching pages as a cognitive state transition network, wherein each cognitive state node in the cognitive state transition network represents the cognitive state at a certain moment in the teaching process. The path optimization module is used to obtain the cognitive transition cost by quantifying the cognitive jump difficulty between adjacent cognitive state nodes, and to determine the optimal teaching path from the initial cognitive state to the target cognitive state based on the principle of minimizing cognitive transition cost. The segment mapping module is used to determine the sequence of teaching segments based on the optimal teaching path. Each teaching segment corresponds to a cognitive state node, which is a mapping system based on educational cognition theory to page structure. The parameter adjustment module is used by the system to calculate the segment parameter vector for each teaching segment after determining the corrected segment type, adjust the segment parameters, and determine the segment parameter vector. The content generation and writing module is used to map each teaching section into a page container; The linkage establishment module is used to establish the linkage relationship between the interactive triggering section and the update target section, so that the learner's operation in a single section can trigger content updates across sections. The output module is used to output the teaching pages.

9. An electronic device, characterized in that, The electronic device includes: At least one processor; and, A memory communicatively connected to the at least one processor; wherein, The memory stores computer program instructions that can be executed by the at least one processor to enable the at least one processor to perform the cognitive state transition-based page adaptive construction method as described in any one of claims 1-7.

10. A computer-readable storage medium storing a program, characterized in that, When the program is executed by the processor, it implements the page adaptive construction method based on cognitive state transition as described in any one of claims 1-7.

Images